qi talk time - hse.ie...– special cause of variation – assignable cause – unstable process –...
TRANSCRIPT
API, 2018
Measurement for Improvement Lloyd Provost 3rd July 2018
Connect Improve Innovate
Building an Irish Network of Quality Improvers QI TALK TIME
API, 2018
Speaker Lloyd P Provost:
An improvement advisor with Associates in Process Improvement who has worked worldwide to apply the science of improvement. His experience includes consulting and teaching in collaborative improvement, planning, management systems, measurement, planned experimentation, and statistical process control. Lloyd is also senior fellow of the Institute of Healthcare Improvement (IHI). Lloyd also works with IHI on their “Improvement Advisor Development Program”.
Lloyd holds has a Bachelor of Science in Statistics from the University of Tennessee and a Master of Science in Statistics from the University of Florida. He is the author of multiple papers relating to quality and measurement and co-author of three books on planned improvement including the Health Care Data Guide (Jossey-Bass, 2011).
API, 2018
Instructions • Interactive
• Sound:
Computer or dial in:
Telephone no: 01-5260058
Event number:840 185 291 #
• Chat box function
– Comments/Ideas
– Questions
• Keep the questions coming
• Twitter: @QITalktime
API, 2018
Lloyd P. Provost Associates in Process Improvement
Measurement for Improvement
3 July, 2018
API, 2018
Webinar Objectives
• Review the key ideas on measuring for improvement.
• Appreciate the importance of analysing measures using time series charts.
• Appreciate learning from special causes.
• Review some examples of the useful types of control charts in healthcare applications.
5
Reference: The Health Care Data Guide, Provost and Murray, 2012
MEASUREMENT FOR IMPROVEMENT
Project Measures: Overall results related to the project
aim (outcome and process
measures). Also Balancing
measures
PDSA Measures Quantitative data on the impact of a
particular change
Qualitative data to help refine the change
Subsets or stratification of project
measures for particular patients
Whatarewetryingtoaccomplish?
Howwillweknowthatachangeisanimprovement?
Whatchangecanwemakethatwillresultinimprovement?
ModelforImprovement
Act Plan
Study Do
From:TheImprovementGuide,AssociatesinProcessImprovement
Aim
Measures
Ideas
Act Plan
Study Do
19 6
API, 2018
March, 1997 The Joint Commission
Journal on Quality Improvement,
Vol 23, No 3.
We are increasingly realizing not
only how critical measurement is to
the quality improvement we seek
but also how counterproductive it
can be to mix measurement for
accountability or research with
measurement for improvement.
7
API, 2018
Percent of Patients Counseled on Smoking Cessation
0
20
40
60
80
100
Jan
06
M M J S N J-
07
M M
%
Percent of Smokers Who Have Not Smoked for Two
Months
0
10
20
30
40
50
60
Jan
06
M M J S N J-
07
M M
%
ArticleFree Gum or Patch
Support GroupE-mail BuddiesArticle
Free Gum or PatchSupport Group
E-mail Buddies
Data for Judgment vs. Improvement
HC Data Guide, p 29, 30 8
API, 2018
Data for Judgment vs. Improvement
Ave Patient Satisfaction Scores
60
70
80
90
100
Jan
06
M M J S N J-
07
M M
Percen
t
Scripting Rec Process Test w ait chg
Patient Satisfaction Percentile Ranking
60
70
80
90
100
Jan
06
M M J S N J-
07
M M
Percen
tile
Scripting Rec Process Test w ait chg
HC Data Guide, p 31 9
API, 2018
Guidelines for Collecting Data for Improvement
• A few key measures that clarify the aim of the improvement effort and make it tangible should be regularly reported throughout the life of the project.
• Be careful about over-doing process measures. A balance of outcome, process and balancing measures is important.
• Plot data visually on the key measures over time. •
• Make use of existing databases and data already collected for developing measures.
• Whenever feasible, integrate data collection for measurement into the daily work routine.
HC Data Guide, p 32 10
API, 2018
Three Categories of Measures
Outcome Measures: Voice of the customer or patient. How is the system performing? What is the result?
Process Measures: Voice of the workings of the system. Are the parts/steps in the system performing as planned?
Balancing Measures: Looking at a system from different directions/dimensions. What happened to the system as we improved the outcome and process measures? (e.g. unanticipated consequences, other factors influencing outcome)
HC Data Guide, p 36 11
API, 2018
Family of Measures for Improvement Project
Surgical Infection Rate
0.000
0.010
0.020
0.030
0.040
0.050
Mar A M J J A S O N D J
Rate
Median 0.026
Patient Satisfaction
75
80
85
90
95
100
Mar A M J J A S O N D J
Perc
ent V
ery
Good/E
x
Median 89
Average Cost per Case
6.4
6.6
6.8
7
7.2
7.4
7.6
7.8
Mar A M J J A S O N D J
Dolla
rs (
k) Median 7.2
Staff With SafetyCulture Score >4
0
10
20
30
40
50
60
70
80
90
100
Mar A M J J A S O N D J
Perc
ent
Median 44
Percent On-Time Antibiotic Use
40
50
60
70
80
90
100
Mar A M J J A S O N D J
Perc
ent
Percent Appropriate Antibiotic Selection
40
50
60
70
80
90
100
Mar A M J J A S O N D J
Perc
ent
Median 86.2
Median 89.9
Figure 2.27: Surgical Safety Family of Measures HC Data Guide, p 61-64
Process Measure
Process Measure
Process Measure
Balancing Measure
Balancing Measure
Outcome Measure
12
API, 2018
Monthly Measure – Goal = 90%
Month Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov
Measure 83 80 81 84 83 85 68 87 89 92 91
60
65
70
75
80
85
90
95
100
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar
Pe
rce
nt
Run Chart of Measure
Goal
Why a run chart? Why not just a table?
HC Data Guide, p 68 13
API, 2018
Average Before=8 hours delay Average After=3 hours delay
Need for Run Charts on Improvement Projects
HC Data Guide, p 16-17 14
API, 2018
QI Tools to Learn
from Variation in Data
HC Data Guide, p 65 15
160
180
200
220
240
260
280
300
320
LO
S (
min
ute
s)
Goal
Work-up done on floor
Bed ahead
Individual responsible
for bed control
Quick-look x-
rays
2/16/98 3/16 4/13 5/11 6/8
Week
Minimum Standard for QI Studies: Annotated Time Series for Outcome and Process Measures
API, 2018
Shewhart’s Theory of Variation
• Common Causes—those causes inherent in the system over time, affect everyone working in the system, and affect all outcomes of the system – Common cause of variation – Chance cause – Stable process – Process in statistical control
• Special Causes—those causes not part of the system all the time or do not affect everyone, but arise because of specific circumstances – Special cause of variation – Assignable cause – Unstable process – Process not in statistical control
HC Data Guide, p. 108 17
API, 2018
Special Causes—those causes not
part of the system all the time or do
not affect everyone, but arise because of specific
circumstances
API, 2018
Shewhart Charts
The Shewhart chart is a statistical tool used to distinguish between variation in a measure due to common causes and variation due to special causes
(Most common name is a control chart, more descriptive would be learning charts or system performance charts)
HC Data Guide, p. 113 19
API, 2018
Why Not Use a Run Chart for Everything?
HC Data Guide, p. 118, 119 20
API, 2018
• Selection of a measure and a statistic to be plotted.
• A method of data collection: observation,
measurement and sampling procedures.
• A strategy for determining subgroups of
measurements (including subgroup size and
frequency).
• Selection of the appropriate Shewhart chart.
• Criteria for identifying a signal of a special cause.
The Method of Shewhart Charts
HC Data Guide p. 114 21
API, 2018
Using Shewhart Chart to Guide
Improvement Work
HC Data Guide, p. 109 22
API, 2018
Learning from Special Causes
23
Conclusion: This case study highlights the utility of statistical
process control in the surveillance and investigation of CA-BSI.
API, 2018
A Special Cause in August
24 26 21 36 34 24 18 24 15 21 27 12 6 6 7 9 9 9 5 8 5 3 8 6 5 8 8 6 6 5 5 4 4 1 4 4 4 9
8340
8741
8546
9034
9080
8890
9324
10054
10450
10667
11007
10491
3861
4094
3827
3625
3706
3434
3439
3790
3757
3651
3705
3840
4180
4484
4177
4159
4353
4306
4545
4326
4834
4476
4595
4483
4620
4874Device
Days
Infections
CCHMC Central Venous Catheter (CVC) Associated
Laboratory Confirmed Bloodstream Infections (LCBIs)
0.2
1.8
0.9
0.8
0.90.9
1.9
2.7
1.11.41.4
0.8
2.2
2.12.4
2.5
1.5
1.8
1.2
1.6
1.31.5
2.6
1.81.6
1.1
2.5
2.0
1.4
2.4
1.9
3.7
4.0
2.5
3.0
2.9
1.2 0.9
0
1
2
3
4
5
6
7
8
Q3 '04
Q4 '04
Q1 '05
Q2 '05
Q3 '05
Q4 '05
Q1 '06
Q2 '06
Q3 '06
Q4 '06
Q1 '07
Q2 '07
Ju
l '0
7
Au
g '07
Sep
'07
Oct
'07
No
v '07
Dec '07
Jan
'08
Feb
'08
Mar
'08
Ap
r '0
8
May '08
Ju
n '08
Ju
l '0
8
Au
g '08
Sep
'08
Oct
'08
No
v '08
Dec '08
Jan
'09
Feb
'09
Mar
'09
Ap
r '0
9
May '09
Ju
n '09
Ju
l '0
9
Au
g '09
Sep
'09
Oct
'09
No
v '09
Dec '09
Jan
'10
Feb
'10
Mar
'10
Ap
r '1
0
May '10
Ju
n '10
FY2005 FY2006 FY2007 FY2008 FY2009 FY2010
Infe
cti
on
s p
er
10
00
De
vic
e D
ay
s
CVC-LCBIs Control Limits Goals [0.8 (Jul06-Jun07) / 1.0 (Jul07-Jun08) ] Baselines
Q3/04 - Revised Care Practices and
Q3/04 - CHG Scrub for Line Care
Q3/04 - CHG Scrub for CVC Insertions
Q4/04 - Maximal sterile barriers and CHG
Q4/04 - Interventional Radiology
New CCHMC
Collaborative
Began, Q1/06
CA-BSI NACHRI
derived CVC bundle
rolled out to hospital
3/15/07
Q2/05 - MaxPlus Cap in PICU B4 & A6
Q2/05 - MaxPlus Cap on A5N2
Q3/05 - MaxPlus Cap cancelled on A5N2
Q3/05 - MaxPlus Caps cancelled.
Updated Thru 31Aug09 by Art Wheeler, Legal/HPCE Depts. Source: Infection Control Dept.
Desired
Direction of
Change
Chart Type: u-chart
3/17/09 - Microclave
Cap Use Housewide
Jan09 - Microclave
Cap Use ICUs
Feb09 - Microclave Cap Use
HemOnc/BMT
From Derek Wheeler, CCHMC 24
API, 2018
Pareto Chart
of Isolated
pathogen Stratified by
Common Cause
and Special
Cause Periods
From Derek Wheeler, CCHMC
Compare Special Cause Period with Common Cause
25
API, 2018
CCHMC Central Venous Catheter (CVC) Associated
Laboratory Confirmed Bloodstream Infections (LCBIs)
0.2
1.8
1.10.9 1.0
0.91.2
2.9
3.0
2.5
4.0
3.7
1.9
2.4
1.4
2.0
2.5
1.1
1.6 1.8
2.6
1.51.3
1.6
1.2
1.8
1.5
2.5
2.4 2.1
2.2
0.8
1.4 1.4 1.1
2.7
1.9
0.9 0.9
0.8
0.9
0
1
2
3
4
5
6
7
8Q
3 '04
Q4 '04
Q1 '05
Q2 '05
Q3 '05
Q4 '05
Q1 '06
Q2 '06
Q3 '06
Q4 '06
Q1 '07
Q2 '07
Ju
l '0
7
Au
g '07
Sep
'07
Oct
'07
No
v '07
Dec '07
Jan
'08
Feb
'08
Mar
'08
Ap
r '0
8
May '08
Ju
n '08
Ju
l '0
8
Au
g '08
Sep
'08
Oct
'08
No
v '08
Dec '08
Jan
'09
Feb
'09
Mar
'09
Ap
r '0
9
May '09
Ju
n '09
Ju
l '0
9
Au
g '09
Sep
'09
Oct
'09
No
v '09
Dec '09
Jan
'10
Feb
'10
Mar
'10
Ap
r '1
0
May '10
FY2005 FY2006 FY2007 FY2008 FY2009 FY2010
Infe
cti
on
s p
er
10
00
De
vic
e D
ay
s
CVC-LCBIs Control Limits Goals [0.8 (Jul06-Jun07) / 1.0 (Jul07-Jun08) ] Baselines
Q3/04 - Revised Care Practices and CHG
Scrub for Line Care
Q3/04 - CHG Scrub for CVC Insertions
Q4/04 - Maximal sterile barriers and CHG
Q4/04 - Interventional Radiology
New CCHMC
Collaborative
Began, Q1/06
CA-BSI NACHRI
derived CVC bundle
rolled out to hospital
3/15/07
Q2/05 - MaxPlus Cap in PICU B4 & A6
Q2/05 - MaxPlus Cap on A5N2
Q3/05 - MaxPlus Cap cancelled on
A5N2
Q3/05 - MaxPlus Caps cancelled.
Desired
Direction of
Change
Chart Type: u-chart
3/17/09 - Microclave
Cap Use Housewide
Jan09 - Microclave
Cap Use ICUs
Feb09 - Microclave Cap
Use HemOnc/BMT
Spiros device
evaluated 8/1/09-
9/2/09
From Derek Wheeler, CCHMC
API, 2018
Rules or determining a special cause
HC Data Guide p. 116 27
API, 2018
#
of
Ne
ed
les
tic
ks
Employee Needlesticksc c ha r t
UCL = 12.60
Mean = 5.54
New Needles Test
1-05 3-05 5-05 7-05 9-05 11-05 1-06 3-06 5-06 7-06 9-06 11-06 1-07 2-07
0
5
10
15
20
28
API, 2018 29
Identify the signals of special causes on these 4 Shewhart Charts
API, 2018
Chart 1
30
58.3
75.4
LCL 41.3
30.0
40.0
50.0
60.0
70.0
80.0
90.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Tim
e -
min
ute
s
Week
Time - Chart 1 Measure
API, 2018 31
61.4
UCL 75.4
LCL 47.3
30.0
40.0
50.0
60.0
70.0
80.0
90.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Tim
e -
min
ute
s
Week
Time - Chart 2 Measure
Chart 2
API, 2018 32
61.2
UCL 71.9
LCL 50.5
40.0
45.0
50.0
55.0
60.0
65.0
70.0
75.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Tim
e-
Min
ute
s
Week
Time - Chart 3 Measure
Chart 3
API, 2018
Chart 4
33
60.5
UCL 70.1
LCL 51.0
40.0
45.0
50.0
55.0
60.0
65.0
70.0
75.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Tim
e (
min
ute
s)
Week
Time - Chart 4 Measure
All common cause variation
API, 2018 HC Data Guide p. 151 34
API, 2018
Continuous (aka variables) Data
• Examples of Continuous Data: – Waiting time – LOS – Cost per case for a DRG – Daily patient weight – Time to complete procedure – Number of patients seen per day – Monthly accounts receivable – Temperature – Volume of prescriptions filled
35
API, 2018
Attribute Data Count or Classification Data
Examples of Count or Classification Data:
– # of complications per # of surgeries this month
– # of medication errors per 1000 doses
– Percent of patients who were readmitted
– # of patients who fell per 100 admissions
– Percent of diabetic patients who smoke
36
API, 2018
Control Chart for Continuous Data using Individual Measurements(I Chart)
The I chart for individual data is useful when: – There is no rational way to organize
the data into subgroups.
– Measures of performance of the process can only be obtained infrequently.
– The variation at any one time (within a subgroup) is insignificant relative to the between subgroup variation.
Examples: patient-specific clinical measures, monthly accounting data, laboratory test data, forecasts, and budget variances
HC Data Guide, p. 154 37
API, 2018 38
Individual Patient Number order they presented to ED
Harries, Filiatrault, and Abu-Laban, “Application of quality improvement analytic methodology in
emergency medicine research: A comparative evaluation”, CJEM 2018:1–8.
Example of an I chart A study of Ottawa ankle rules (OAR) by
emergency department (ED) triage nurses.
API, 2018
X-bar and S Shewhart charts
• The collection of data for the construction of X-bar and S Shewhart charts requires that the data be organized in subgroups.
• A subgroup for continuous data is a set of measurements of some characteristic in a process, which were obtained under similar conditions or during the same time period • The subgroup size may vary for the X-bar and S chart. • The X-bar chart contains the averages of each subgroup and the S chart the spread (standard deviation) between the measurements within each subgroup.
HC Data Guide, p. 160 39
API, 2018
Xbar and S Chart- Radiology Test Turn-around
HC Data Guide, p. 160-161
June 7 June
8
June 9 June
10
June
11
June
12
June
13
June
14
June
15
June
16
June
17
June
18
June
19
June
20 Test 1 105 50 58 67 73 57 78 76 18 39 86 32 70 39 Test 2
54 105 26 52 59 64 96 49 60 45 27 47 56 34 Test 3
79 85 38 92 62 24 107 40 39 35 91 57 40 146 Test 4
49 46 72 49 34 Test 5
31 27 23 Test 6
43 65
40
API, 2018 41 “Improving Adherence to Intraoperative Lung-Protective Ventilation Strategies at a University
Medical Center” Josephs, Lemmink, Strong, Barry, Hurford, www.anesthesia-analgesia.org
Intraoperative lung-protective
ventilation (ILPV) is defined
as tidal volumes <8 mL/kg
ideal bodyweight and is
increasingly a standard of
care for major abdominal
surgical procedures
performed under general
anesthesia.
Subgroups
of patients
based on
time periods
Example
Xbar/S Chart
API, 2018
Xbar S – Subgroups by Provider and Quarter D
ays
Average LOS by Provider for DRG XXX
3.75 4.00 3.62 4.35 3.79 4.12 3.71 3.93 3.62 3.81 3.68 3.50 5.18 4.65 4.27 3.95 3.89 3.75 4.13 4.42Ave LOS
Xbar chart
Sigma chart
UCL = 4.69
Mean = 4.00
LCL = 3.32
UCL = 1.93
Mean = 1.45
LCL = 0.96
21/Q1 Q2 73/Q1 Q2 35/Q1 Q2 18/Q1 Q2 10/Q1 Q2 31/Q1 Q2 66/Q1 Q2 45/Q1 Q2 89/Q1 Q2 17/Q1 Q2
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
0.5
1.0
1.5
2.0
2.5
HC Data Guide, p. 162 42
API, 2018 HC Data Guide p. 151 43
API, 2018
Shewhart Charts for Attribute Data
Classifications of units: each unit is classified as either conforming or
nonconforming, pass/fail, blue/not blue, go/no-go, etc.
Count of incidence: a count of the number of nonconformities, defects
(complications, infections, errors), accidents, trips, telephone calls, etc.
Chart Type of Subgroup
Name Attribute Data Statistic Charted Size
P Chart classification % nonconforming units (P) constant or may vary
C Chart count number of incidence (C) constant
U Chart count incidents per unit (U) may vary
Table 5.4: Three Types of Attribute Shewhart Charts
HC Data Guide, p. 163 44
API, 2018
P- Charts Example
45 Zafar MA, et al. BMJ Qual Saf 2017;26:908–918. doi:10.1136/bmjqs-2017-006529
API, 2018
P chart as Funnel Plot
P chart for MRSA Subgroups are Hospitals
46
API, 2018
C- Chart #
of N
ee
dle
stic
ks
Employee Needlesticksc chart
UCL = 12.60
Mean = 5.54
New Needles Test
1-05 3-05 5-05 7-05 9-05 11-05 1-06 3-06 5-06 7-06 9-06 11-06 1-07 2-07
0
5
10
15
20
HC Data Guide, p. 177-178 47
API, 2018
Figure 1
Example U Chart
API, 2018 HC Data Guide p. 220 49
API, 2018
“Cases Between” Occurrences of Rare Events
Instead of plotting the
number of incidences
each month, plot the time
(or number of cases,
patients, visits, etc.)
between incidences.
Plot a point each time an
incidence occurs
HC Data Guide: p. 227
API, 2018
g-chart for Doses Dispenses between ADE’s
HC Data Guide: p. 228, 230
gbar = average of g’s (units between events)
CL = 0.693 * gbar (estimate of median for geometric distribution)
UL = gbar + 3 * square root [gbar * (gbar +1)] 51
API, 2018
T chart Showing Improvement
HC Data Guide: p. 233 52
API, 2018
Establishing and Revising Limits for a Shewhart Charts
HC Data Guide p. 122
Update Trial Limits
Start with a Run Chart
Calculate Trial Limits
53
API, 2018 HC Data Guide p. 119 54
API, 2018
HC Data Guide p. 121 55
API, 2018
Updating Limits
56
API, 2018
Deming, W. Edwards: Foreword in
Statistical Method from the Viewpoint of
Quality Control. Dover Publications,
1986, p. ii. ONLY 18 MORE YEARS!
57
API, 2018
Helpful links Framework for Improving quality
www.qualityimprovement.ie Improvement Knowledge and Skills Guide
http://www.hse.ie/eng/about/Who/QID/aboutQID/
API, 2018
Thank you from all the team @QITalktime
Follow us on Twitter @QITalktime
Watch recorded webinars on HSEQID
QITalktime page
Please give us some feedback on survey monkey:
https://www.surveymonkey.com/r/8WNQVH9
We will be back with some exciting events post
summer.